AI: a question of economy?

Because it is a total social fact, the rapid emergence of artificial intelligence must be considered from all angles, so that it benefits the economy without being solely determined by it. We are on the threshold of a new paradigm.

Fabien Seraidarian
Director of Knwoledge Transfer and of the Global Executive MBA, SKEMA Business School

“Football is a total social fact because it concerns more or less every aspect of society, and also because it can be seen from different points of view,” wrote anthropologist Marc Augé in Le Monde diplomatique (1998). But isn’t the same true for artificial intelligence (AI)? Beyond the technological challenges, AI is transforming the economy, calling our sovereignty and institutions into question, and influencing our social and cultural habits right down to our daily lives.

The concept of the “total social fact”, first a basis for Emile Durkheim’s sociology and then a marker in the work of Marcel Mauss, suggests an anthropological approach, so as not to reduce artificial intelligence to an artefact, i.e. a digital product resulting from automatic learning or a symbolic AI. Adopting an interdisciplinary approach is becoming a necessity, but we also need to take account of “where we are talking” in order to understand opinions, representations, ideologies and the many intercultural issues at stake.

Marc Augé’s approach is also inspiring for exploring the impact of AI and the advent of a “super modernity”, which he feels stands apart from modernity through three essential aspects: a time overloaded with events; the accelerated transformation of the contemporary world and the geographical shrinking of the global world, and lastly the individualisation of references. ”Never (…) have the benchmarks of collective identification been so fluid”; “the individual production of meaning is now more vital than ever.” The major characteristic of supermodernity seems to be the excesses that can now be abundantly nourished by artificial intelligence.

In practice, there is a great temptation to be techno-sceptical (tech for worst) or pro-technology (tech for good), and to reduce the complexity of this massive innovation that is invading the social, private and professional spheres. We have adopted an approach that connects economic impact, the transformation of firms, individuals and skills development, and the societal perspective (including the role of institutions and cultural dynamics). Each of these dimensions provides an up-to-date (analytical) view of AI to assess its performance. But our reality is like a (complex) system that makes arbitration difficult, as in the recent debates on the AI Act, which show how difficult it is to reconcile the interests of citizens, companies, States and Europe.

Companies need men and women leaders, each bringing their own particular vision and added value.

AN ECONOMIC IMPACT THAT HAS YET TO BE SEEN…

Economic growth stems more from technical progress than from the accumulation of a stock of capital. Information and communication technologies (ICTs) have been the focus of much attention since the late 1990s and have failed to live up to their promise of productivity gains: “We see computers everywhere, except in productivity statistics”, as the economist Robert Solow said in 1987. Subsequently, endogenous growth theories demonstrated that it is intangible investment in R&D and human capital that generates positive externalities and ensures long-term growth.

It still seems too early to empirically assess the effect of AI on growth, but initial microeconomic studies suggest significant positive effects on the productivity of individual workers with some of its specific applications. Productivity gains concern the “least productive” employees, thus helping them catch up with the “most productive”. However, the effects of AI on business productivity that have been measured are still limited: AI’s adoption is still infrequent and uneven within companies, although it is clearly more prevalent in large companies and digital firms.

AI’s effects on employment are also cause for considerable concern: they remain uncertain and depend on deployment capacities, changes in professions and the reallocation of the workforce. The effects of AI on aggregate labour demand will depend on the processes already observed during previous technological breakthroughs, i.e. the “creative destruction” mechanisms defined by Schumpeter. Empirical studies show that AI is more likely to affect skilled occupations where more abstract and non-routine tasks are handled. But skilled professions would also be the most likely to benefit from the productivity gains enabled by the adoption of AI. This brings us full circle to the machine vs. tool dialectic discussed above.

However, there is consensus on the issues of education and the need to bolster training, from primary to higher education, as well as for continuous training in the professions undergoing transformation. This raises the question of access to artificial intelligence and how to train people to use it as a lever for growth. This is one of the ways to develop investment that is very inadequate in Europe and is thus slowing down the digitisation of the economy, particularly in the absence of European leadership.

ARTIFICIAL INTELLIGENCE: A PRODUCTIVITY LEVER FOR COMPANIES?

AI is transforming the relationship between people and technology: advances are not only enabling automation but also greater autonomy in production systems and supply chains. AI reduces the cost of operating processes and improves the ergonomics and safety of workstations. According to Jeremy Rifkin (2013), AI promises marginal costs in production, delivery, (remote) maintenance, control and communication, which “are moving towards 0”. Finally, AI can lead companies to develop increasing yields, representing a radical break with business models: the average cost falls, which implies that marginal costs also fall, while being lower than the average cost. This is a major breakthrough in economic models, which is disrupting the established order in economic sectors.

Many questions remain unanswered, and the years ahead will be full of informative lessons. AI could intensify the dematerialisation and disintermediation of production processes and commercial exchanges. It should also shorten value chains and decision-making circuits within organisations and at the ecosystem level. Lastly, AI could encourage the emergence of new, more agile and effective forms of open innovation, ranging from R&D (living labs, fablabs, etc.) to cooperative production (digital microfabrication, do-it-yourself, makerspace, etc.) and collaborative consumption (hosting, peer-to-peer, car-sharing, etc.).

Whether healthcare (diagnostic assistance, disease prediction, robotic surgery), the automotive industry (smart cars, predictive maintenance, etc.), financial services (autonomised relationships and transactions), distribution (customised design, automated management, etc.), manufacturing (optimised supply chains), logistics (automated deliveries) or energy (optimised operation, smart maintenance, etc.), every sector seems eligible for what AI is offering. AI is also leading to new risks, including cybercrime, which requires the development of new tools to prevent these risks occurring for companies and their customers.

AUGMENTED EMPLOYEE OR TRANSHUMANISM?

When could AI finally be able to rival and even surpass human faculties? Should we be worried? As individuals, we are employees and citizens in turn. AI is a massive innovation and is not confined to the economic sphere like robotics, which has transformed production lines. The numerous interviews carried out here show that two uses are becoming established in companies: AI serving the performance of the company’s activities, and AI as a personal assistant for employees, whose capabilities and productivity will thus increase. When the machine/ tool dialectic leans towards AI being an “extension of the human hand” and ethical rules are established (with the creation and use of databases as a priority), employees can benefit from the potential offered by artificial intelligence applications.

But is it that simple? A historical perspective provides some nuances on the uses of AI for employees. After giving rise to a host of fantasies and fears in recent years, AI is now becoming established in practice. And in many respects, the interviews here shed light on practices: companies are seizing on applications in the market to customise them and integrate them into their technological environment. Projects are springing up everywhere, from in-house versions of ChatGPT to various business applications designed to make employees’ daily lives easier. After a period when teams of data scientists were developing prototypes to demonstrate AI’s virtues, business lines are now proposing to rethink their practices, auguring a growing appropriation of artificial intelligence.

Many questions remain unanswered. By definition, it seems likely that over time the majority of employees will have to reinvent their jobs and activities if AI can develop applications to handle multiple tasks and replicate know-how. In 2018, McKinsey predicted that AI-based technologies would rapidly increase the efficiency of work by 20% to 40% depending on the industrial country. Businesses, and society in general, will be facing major challenges in the coming years, which will involve not only education, training and skills development, but also job security in the sectors and professions most affected. Otherwise, AI will widen the inequalities between those who make it a tool to greatly increase their capabilities and those who end up impoverished by the machine.

FROM ONE PARADIGM TO ANOTHER?

AI is making lightning progress and disrupting society. This dynamic is far too fast to accommodate social realities, and could cause a growing gap between individuals (professional identity, income, etc.) and the players who build AI systems and create new standards. A paradigm shift is needed to create a system.

While there are intense debates between those who want minimal regulation to foster the emergence of national champions, and those advocating regulation to prevent the risks generated by AI, the role of national institutions and public policies, and further afield, is a major issue to be addressed if society is not to be split.

The development of AI raises the question of the distribution of the commons, long advocated by the French economist Gaël Giraud. To put it simply, the idea is to share the management of common resources. These are the conclusions of Jeremy Rifkin who, in his 2014 book The Zero Marginal Cost Society, imagined an economy of interdependencies and “collaborative commons” based on the pursuit of a community’s interests rather than individual aspirations.